OpenAI’s goal of going public in the fourth quarter of 2026 now appears unattainable, according to a new analysis from PitchBook. The company’s heavy financial commitments are forcing a more realistic target of mid-to-late 2027. PitchBook added that public investors will demand several additional quarters of steady performance to understand how more than $1.15 trillion in long-term infrastructure deals can generate meaningful free cash flow.
This delay reshapes the competitive landscape for frontier AI companies preparing to list. At the center of the issue is a fundamental imbalance.
The research report and analysis from PitchBook indicated that OpenAI currently brings in about $2 billion in monthly revenue—an impressive figure by traditional standards.
Yet it has locked in fixed, non-negotiable spending commitments totaling $1.15 trillion with partners including Oracle, Microsoft Azure, Amazon Web Services, NVIDIA, AMD, Broadcom, and CoreWeave.
The Oracle agreement alone calls for $60 billion annually starting in 2027, surpassing the firm’s own projected net revenue for that year.
These obligations cannot shrink even if growth slows, creating a dangerous mismatch when revenue targets are missed, as happened earlier this year in coding and enterprise segments where Anthropic gained ground.
A joint statement from CEO Sam Altman and CFO Sarah Friar dismissing reports of internal disagreements over compute spending proved more revealing than the revenue shortfall itself.
PitchBook further noted that publicly addressing such tensions at this pre-IPO stage signals deeper strategic friction around capital allocation, especially as Friar has raised legitimate concerns about funding future contracts without accelerated growth.
The gap with rivals is widening. Anthropic operates with roughly one-twelfth of OpenAI’s infrastructure burden, enjoys stronger gross margins, and is expanding faster in key areas OpenAI has lost.
Revenue efficiency highlights the divergence: Anthropic produces approximately $6 million in annualized revenue per employee across a 5,000-person workforce, while OpenAI generates about $5.6 million per employee with 4,500 staff and plans to nearly double head count by year-end.
Anthropic is improving efficiency as it scales; OpenAI is layering costs onto an already strained structure.
PitchBook’s AI Business Quality (AIBQ) scoring framework shows pressure mounting simultaneously across three areas.
Governance optionality—already OpenAI’s weakest metric at 3 out of 10—faces fresh scrutiny from the public leadership disagreement.
Revenue quality is under threat if market share losses persist, and capital efficiency is deteriorating as new commitments, such as an additional $100 billion from AWS, accumulate against more than $180 billion already invested.
Maintaining the current 4.2 composite score looks increasingly difficult.
According to the insights from PitchBook, the broader stakes are clear. The first frontier AI company to reach the public markets will set the valuation benchmark for the sector.
If Anthropic or Databricks lists ahead on cleaner economics, OpenAI risks entering a valuation framework it did not shape and cannot easily influence, despite having deployed the most capital. PitchBook concluded that revenue fluctuations may fade quickly, but the rigid cost base will not. For OpenAI, the real price of further delay could be ceding control over how the market ultimately values the entire industry.